For the first time, wireless sensor networks are on the brink of becoming part of everyday life. With growing relevance, security risks associated with pervasively communicating devices are also increasing. To counter these security threats, both cryptographic and non-cryptographic security measures must be applied in cooperation with each other. Security threats that cannot be counteracted with cryptographic means are usually accompanied by specific, measurable changes in networking characteristics such as signal strength, message loss, etc. This work explores the possibility to detect network traffic anomalies using a lightweight statistical model. It allows the sensor nodes in the network to locally observe and learn typical networking characteristics. Deviations from the expected behaviour are detected and classified as anomalous.
Denise Miriam Dudek
Cryptography Lightweight Anomaly Detection Wireless Sensor Networks